12 Feet Under: 1000 Square Feet of Geothermal PC Cooling

[Romir]

Stock 920, 1.15v load. The is update removed my profiles. I'm going to run the OCCT GPU+CPU test overnight but the ambient air will cool the exposed pipes and skew the results downwards.

[=[PULSAR]=]

Nice results thus far, any issues with condensation yet?

[Romir]

Nice results thus far, any issues with condensation yet?

Its not a problem in the winter with low humidity. The water was 47f last I checked and its 60f in here with 30% humidity. Brrr.

Using the heat exchanger and a slower, quieter pump should keep my temperatures high enough in the summer, if needed. A de-humidifier is definitely planned.



The Iwaki is using 80w now, suggesting a lighter load without the 3 PC blocks in the loop (plate heat exchanger instead). The 500 feet of added piping today was no problem for it.

[Romir]

Here are 100 half-sized photos of the dig.

http://picasaweb.google.com/11578270...opInstallation


The main thing I need to note is how hitting the water pipe resulted in the far end of the trench being too short. Basically we could only dig one way once we ran into that parallel pipe.

That resulted in the distorted, shallower, and stretched piping at that end. I'm sending the warmed water down to that end first, that way the water will exit through the deepest and coldest coil section at the other end.

[Romir]

The weather was absolutely perfect today. I've now finished 99% of everything that needs to be done outside. The loop is essentially finalized from a performance perspective. It already seems safe to say the loop temperature won't be affected by my computer unless its doing something 24/7 like folding.

My goals this week are finalizing the indoor plumbing and getting the 1-wire temperature graphing going.



5/8th wall thickness insulation on the return piping. The trench runs downhill so I'm not concerned about moisture collecting inside the insulation.



The money shot. My grossly over-sized radiator, hidden out of sight.



The pvc valves need to be angled towards to the ground pipes for a straight hose connection. This picture doesn't show that well unfortunately. A 45 connector should do it for the left valve. The right one can be cut, rotated, and rejoined with a coupler.



The heat exchanger between my PC loop and the geothermal loop. I bought another one for the 2nd PC, or to test if dumping room heat into radiators can make a measurable difference. The temperature gun isn't cooperative on this shiny metallic surface. An in-line temperature probe on the PC loop would be useful...



After 90 minutes of bleeding:



The ground probe results. For reference, it's 5 feet down at the start of the geothermal trench. It took less than 30 seconds to slide the conduit pipe into the disturbed soil. I didn't end up burying the ($$$) moisture sensor.



I've given up on trying to raise the water temperature with my PC. I'm only managing to throw money away heating the outside ground. The gpu temperatures remained unchanged after 30m with a 950w measured load at the wall.

Clearly the 97% efficient 1500w heat stick is going to be required. Starting Monday I'll do an extended thermal conductivity test of the loop with it. The results can linearly graphed to determine how much heat it takes to raise the temperature in a given period of time. (1 degree Fahrenheit in 1 hour is the standard measurement.)

It could take all week to do the test because the heat load is undersized for 300' of piping. It'll be cheaper to do a longer test at $3.50 in electricity a day than to assemble another heat stick or two and in-line reservoirs for them.

[voigts]

So to get this straight in my mind, you have decided to make the geothermal loop separate from the actual PC loop and use a heat exchanger between the two? And you are doing so because of concerns over condensation, correct?

I will be curious to see how your testing numbers come out. This really looks like a complicated project.

[Romir]

So to get this straight in my mind, you have decided to make the geothermal loop separate from the actual PC loop and use a heat exchanger between the two? And you are doing so because of concerns over condensation, correct?

I will be curious to see how your testing numbers come out. This really looks like a complicated project.

That's right. Having chilled blocks even when the pc is off was very off-putting. Achieving safe slightly below ambient temperatures with very little power use was the project's goal.

The Iwaki is using less than the 100w I thought it was going to (79w currently), so I'm still coming out ahead electricity wise. Especially considering how the PC is shut down 95% of the time.

Using heat exchanger(s) also buys me more flow from the 30RZT. Without the PC in the loop, I'm supposed to be about 2/3rds up this curve. I haven't tested with the King flow meter yet, but I suspect the three PC blocks and quick disconnects halved the flow. Of course, it probably flowed faster than the single DDC 3.2 currently does, and without the efficiency loss of the exchanger to boot.



It was only complicated with the amount of small plumbing parts it required. I'm having trouble recording the quantity of everything in the expense log. The Lowe's return pile is also quite impressive.

Originally I was going to run one 100' of 1/2" copper tubing in a trench outside my window. My two DDC 3.2s would've powered it. It was a such a simple plan...

Routing to the utility closet and locating down the hill took an additional 250 feet of 3/4" pipe/tubing. That's where things got messy, and it's iffy if it'll pay off from a pure temperature stand point.

[Movieman]

HMM.....Turn that beast on 24/7, install WCG and join the team here.
machine will run at 100% load and solve your lack of load issues..

[Romir]

HMM.....Turn that beast on 24/7, install WCG and join the team here.
machine will run at 100% load and solve your lack of load issues..

I did 2 million points in F@H before the gpu clients, those hundreds of dollars of electricity costs were enough for me. Now I won't even get the side benefit of winter electrical heating.

Here are the results of a 6 hour OCCT max psu load test. It was pulling 680w from the wall with the i7 @ 4GHz HT 1.25v and the 5970 @ 5870 clocks (1.162v 850/1200).






Looks like a 1 degree Celsius rise from the start to end. Not heating the ground up was the primary design goal, and that's looking good! I'm now almost fully at mercy of the ground temperature.

[McBacker]

That's amazing. Very nice results!

[shazza]

Romir ... the scope of this project is almost more than I can comprehend. It sure looks like you're getting good results, and I admire your plans and execution of this project - just amazing stuff.

[BringerOdeath]

My jaw dropped when i saw this!! Unbelievable work romir!! Only hooking my PC up to some kind of nuclear powerplant rod cooling device would be able to beat your geothermal setup....Or maybe a high altitude baloon with my rads on it with tubs all the way back to my house!! I wonder what kind of pump I would need....

[voigts]

No offense, but this project seems like a lot to go through for a system that is only running 5% of the time.

[Wezly]

Nawwwwww

It's super eco-friendly!

[cubemaster27]

see if this project qualifies for a tax credit

awesome awesome project

[[XC] gomeler]

Nice to see you got it done. Now you need to use the loop for indoor residential cooling during the summer

[Sneaky]

I've seen geothermal loops in the past, but NOTHING like this! that's a hell of an undertaking, but I'm glad to see that you followed through with it and got it finished. Good job!

[SovietKitsch]

The scale of this is just incredible. Great job!

[kgtiger]

Congratulations mate on getting the tubing in the ground.
Now just the fiddly bits and then it's all fun from here on.

[p2501]

Can't describe how epic this project is. People like you, doing projects like this are really an inspiration! Keep the updates coming!

What's the capacity of the HX by the way?! Any chance of overloading that small thing?!

[zads]

Awesome project, subscribed!
Hmm maybe I should use this method with my pool, that thing is way too cold to swim in for 75% of the year anyways :P

[hyperdoggy]

Wow this is awesome, i wish i can do something like that but living in NYC puts it out of the question.

[laevanira417]

Oh...
It's Great project!!

[Duh]

1st of all: my hat off .. very nice

No offense, but this project seems like a lot to go through for a system that is only running 5% of the time.

you are trying to make the point that he should enter to our WCG team?

[DashTrash]

Any updates? How did it perform over the summer?